Research On Micro-seismic Event Detection Algorithm Of Hot Dry Rock Based On Neural Network

The global problem of coal,oil and gas resources is serious,and the development of safe and sustainable clean energy has become the focus of resource exploration.Hot dry rock is a renewable underground resource with temperature above 180 ℃ and extremely low permeability.After detecting dry hot rock resources,energy acquisition for dry hot rock is implemented.and hydraulic fracturing technology is used to create an underground reservoir.Creating an underground reservoir using hydraulic fracturing combined with microseismic monitoring technology.By analyzing the microseismic events generated during hydraulic fracturing.Determine the fracture trend of underground reservoir.Further adjust the hydraulic fracturing plan.In order to obtain accurate underground reservoir information,the detection of microseismic events is a key step.The data collected at the project site contains complex and variable waveforms.Therefore,how to detect microseismic events quickly and accurately is the key technology in the whole process of microseismic monitoring.This paper aims at the monitoring requirements of hydraulic fracturing in the process of energy acquisition in hot dry rock,the research objective of this paper is to detect the maximum amplitude used to calculate the magnitude from a large number of microseismic data and to process the microseismic data in detail.For these two objectives,the work of this paper includes the following four aspects:Firstly,it analyzes the research background and significance,clarifies the positive significance of clean energy development in dry hot rock,introduces the relationship between hydraulic fracturing technology and micro seismic monitoring in dry hot rock,and briefly describes the current situation of micro seismic event detection at home and abroad.This paper compares and analyzes the microseismic event detection methods used by predecessors,starting with traditional methods and machine learning algorithms,and lists the working principles,advantages and disadvantages of these algorithms.Secondly,Then,according to the needs of the project,a microseismic event detection algorithm based on short-term and short-term memory network optimization is designed.It includes two algorithms,the first is an algorithm based on gated recurrent unit and support vector machine.This algorithm can directly pick up the maximum amplitude in the microseismic event.The basic structure principle,training process and parameter selection of gated recurrent unit and support vector machine are introduced.The second is to design an algorithm based on bidirectional long short term memory networks and conditional random fields for detailed processing of microseismic data.Utilize the advantage of bidirectional long short term memory networks in memorizing information to detect P waves and S waves in data with insignificant characteristics.Then,the paper introduces the subject of micro-seismic monitoring of hot dry rock,gives the micro-seismic monitoring scheme in the process of hydraulic fracturing,and proposes an improved load balance relationship between APs for wireless data transmission,and proposes a polling mechanism to maximize throughput and improve data transmission efficiency.Finally,the two algorithms are tested respectively.Different data are used according to different objectives of the algorithm.The algorithm for gated recurrent unit and support vector machine uses the simulated generated rake wavelet and the micro-seismic data collected by ground instruments,and the algorithm for bidirectional long short term memory network and conditional random field directly uses the data collected by cableless seismometers for testing.After a large number of different types of data testing,the results are analyzed and the conclusion is drawn that the algorithm proposed in this paper can effectively process the micro-seismic data.